Though we unfortunately do not have the source of the information, it was found somewhere on the web that EMD has been used to fairly effectively determine the difference between a "fake" signal synthesized to look like natural data and a natural signal of the same type.

Several experiments we conducted support this finding. While several group members looked online for signals on which to use the EMD, one group member attempted to get a meaningful EMD out of signals he created. Several different methods were attempted, including using sine waves whose frequency change with time, with noise added. The decompositions were a mess; we found we could not create a signal with the superposition of causal modes that natural signals so often have.

Hopefully we will soon be able to post some of the created signals mentioned and their EMDs; they have been lost but may yet be rediscovered.